Adult salmon are well known to return to the lake or stream where they hatched, to spawn another generation of salmon. In many places, fisheries catch them in the ocean on the way back to spawning, but before it is possible to assign them to a particular population from a stream or lake. A new model now shows a way forward to disentangling catches that come from multiple salmon populations, using genetics and analysis of scales. The new method combines recent DNA samples with historical DNA obtained from fish scales stored for decades in archives, to assign catches to their natal regions, supplementing more extensive data collected from fish scales that are used to estimate the number of years each salmon spent in freshwater and the ocean. The model is applied to sockeye salmon catches in Bristol Bay, Alaska, which hosts the world’s most largest salmon fishery, but is threatened by a massive proposed mine. The model finds that salmon spending more years at sea before returning, are also more likely to be caught in the fishery, and that there are sometimes large interception rates of salmon heading for one river system but that are caught near the mouth of another river system. The new run reconstruction model estimates large differences in the run sizes of some populations that were formerly thought to be caught in higher or lower numbers. The results will feed back into management estimates of maximum sustainable catch for each Bristol Bay fishing district. A paper describing the method appears in the Canadian Journal of Fisheries and Aquatic Sciences, and is authored by former SAFS graduate students Curry Cunningham and Matt Smith, and SAFS professors Trevor Branch, Lisa Seeb, Jim Seeb, and Ray Hilborn, together with Tyler Dann of the Alaska Department of Fish and Game.